CN105388908A - Machine vision-based unmanned aerial vehicle positioned landing method and system - Google Patents

Abstract

The invention discloses a method and system for positioning and landing of a UAV based on machine vision. The method includes calculating positioning parameters, solving the position parameters of the UAV relative to the landing area, and controlling the positioning and landing of the UAV; The landing system includes a first GPS information receiving unit, which is used to compare the pseudo-range observation value measured by the reference station with the known distance to obtain a pseudo-range correction value; a second GPS information receiving unit, which is used to observe GPS satellite data, receive The correction value from the reference station is used to calculate the positioning parameters; the landing area acquisition unit completes the feature collection of the landing area through the camera fixed on the UAV; the video signal processing unit is used to amplify the video signal output by the CCD; Image processing unit: used to process the characteristic patterns of the photographic landing area; and a control command unit and a landing execution unit. The invention combines DGPS navigation technology and visual navigation technology, so that the drone can land or hover at a fixed point with high precision.

Description

Method and system for positioning and landing of UAV based on machine vision

technical field

The invention relates to the field of unmanned aerial vehicle control, in particular to a method and system for positioning and landing an unmanned aerial vehicle based on machine vision.

Background technique

UAVs are available day and night, simple in structure, easy to use, low in cost, high in efficiency, and do not have to worry about casualties. Therefore, UAV operations are increasingly favored in high-risk environments. It can be used for scene monitoring, meteorological investigation, road inspection, surveying and mapping, flood monitoring, aerial photography, traffic management, forest fires, etc. From this point of view, drones have extremely broad application prospects in many aspects.

In the process of performing missions, fixed-point landing is a very important and prone to failure stage. Research data show that more than one-third of the flight accidents in the world's aviation history occurred during fixed-point landing. Therefore, UAV fixed-point landing technology has become one of the key technologies affecting the development of UAVs. Whether it can realize automatic landing safely and reliably is also an important indicator for evaluating the performance of UAVs.

UAV positioning and landing refers to the process in which the UAV relies on the onboard navigation equipment and flight control system for positioning and navigation, and finally controls the UAV to land on the landing site. In order to achieve fixed-point landing, the UAV must have the ability to navigate autonomously. High-precision fixed-point landing navigation technologies include: inertial navigation system (INS), GPS navigation, INS/GPS integrated navigation system and visual navigation system. Among them, inertial navigation is the earliest and most mature navigation technology; GPS is the most widely used and relatively mature navigation technology that has emerged in recent years. However, the above-mentioned navigation technologies have their own shortcomings. The GPS positioning system is too dependent on navigation satellites; the inertial navigation system uses gyroscopes, accelerometers and other inertial devices to obtain data, and finally calculates the carrier position. is too big. With the emergence of multiple navigation technologies, it is natural to combine different navigation technologies to give full play to their respective advantages to achieve the best navigation effect.

Contents of the invention

The technical problem to be solved by the present invention is to provide a machine vision-based UAV positioning and landing method and system, combining DGPS (DifferenceGlobalPositioningSystem, Differential Global Positioning System) navigation technology with visual navigation technology, so that the UAV can finally Land or hover at a fixed point with high precision.

In order to solve the problems of the technologies described above, the technical solution adopted in the present invention is:

A method for positioning and landing of a UAV based on machine vision, including calculating positioning parameters, solving the position parameters of the UAV relative to the landing area, and controlling the positioning and landing of the UAV;

Calculate the positioning parameters: through two GPS receivers installed on the base station and the UAV, the same GPS satellites are tracked and observed synchronously; the pseudo-range observation value measured by the base station is compared with the known distance, and the pseudo-range correction value is calculated. It is transmitted to the GPS receiver on the drone through the data link; the GPS receiver on the drone observes the GPS satellite data, receives the correction value from the reference station, corrects the observed pseudo-range, and then uses the corrected pseudo-range for positioning. Calculate the positioning parameters;

Solve the position parameters of the UAV relative to the landing area: complete the acquisition of the characteristic characteristics of the landing area through the camera fixed on the UAV; process the collected video signal, that is, the characteristic pattern of the landing area, and the video acquisition card will The video signal output by CCD (Charge-coupled Device, charge-coupled device) is amplified, and the field signal is synchronously separated, and the detected video signal is sent to the image processing circuit; Process to obtain the coordinates of the feature points in the image, and send the coordinates to the main control computer to solve the position parameters of the drone relative to the landing area;

Control the positioning and landing of the UAV: After the main control computer calculates the distance, height and attitude angle of the UAV relative to the landing area, it transmits the data to the UAV; the UAV autopilot system collects The parameters provided by the sensor are used to generate control instructions according to the set control method and logic, and the drone is positioned and landed by controlling the actuator.

According to the above solution, only one camera is provided, and the image capturing pixels are 1920×1080.

According to the above solution, the positioning parameters adopt the WGS-84 coordinate system.

According to the above scheme, the specific method for the image processing circuit to process the characteristic patterns of the landing area of photography is to sequentially perform grayscale, image correction, filtering, feature point extraction, non-feature point exclusion, rough registration, and threshold selection , binarization, feature point extraction, feature point re-detection, linear pose calculation and result output.

A kind of unmanned aerial vehicle positioning landing system based on machine vision, comprises following unit: the first GPS information receiving unit: the GPS receiver of described first GPS information receiving unit is installed on the reference station, is used for measuring the reference station The pseudo-range observation value is compared with the known distance to obtain the pseudo-range correction value, which is transmitted to the GPS receiver installed on the drone through the data link; the second GPS information receiving unit: the GPS receiving unit of the second GPS information receiving unit The machine is installed on the UAV to observe the GPS satellite data, receive the correction value from the reference station, correct the observed pseudo-range, and then use the corrected pseudo-range to perform positioning and calculate the positioning parameters; the landing area acquisition unit: The camera fixed on the UAV completes the acquisition of the characteristic characteristics of the landing area; the video signal processing unit: it is used to amplify the video signal output by the CCD, perform synchronous separation of the field signal, and send the detected video signal to the image processing unit ; Image processing unit: used to process the characteristic patterns of the landing area of photography, obtain the coordinates of the feature points in the image, and send the coordinates to the main control computer to solve the position parameters of the UAV relative to the landing area; the control command unit : Used to calculate the distance, height and attitude angle of the UAV relative to the landing area, and then transmit the data to the UAV; Landing Execution Unit: According to the input control command, collect the parameters provided by the sensor, and follow the set A certain control method and logic are used to generate control instructions, and the drone is positioned and landed by controlling the actuator.

Compared with the prior art, the beneficial effect of the present invention is: the combination of the DGPS navigation system and the visual navigation system enables the UAV to finally land or hover at a fixed point with high precision, and the present invention realizes positioning and landing The functions are faster and more accurate, providing effective support for the better application of UAVs in power line inspections and emergency repairs.

Description of drawings

Fig. 1 is a schematic structural diagram of the UAV positioning and landing system based on machine vision in the present invention.

FIG. 2 is a schematic diagram of the image processing process of the machine vision-based UAV positioning and landing method in the present invention.

Fig. 3 is a schematic flow chart of the machine vision-based UAV positioning and landing method in the present invention.

detailed description

The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. A machine vision-based UAV positioning and landing system provided by the present invention adopts DGPS UAV piloting technology to guide the UAV to the vicinity of the characteristic pattern of the landing area, and start the machine vision assisting system to monitor the UAV. Real-time precise positioning, and the position information of the drone is transmitted to the flight control system, and the flight control system controls the landing of the drone. The invention mainly applies real-time image processing, image feature point extraction and UAV positioning, etc., and is mainly applied to power line fault detection and troubleshooting. The invention mainly includes a DGPS navigation system, a visual navigation system and an unmanned aerial vehicle automatic driving system.

The DGPS of the present invention uses two GPS receivers, one set at a known point (ground control station) as a reference station, one for unmanned aerial vehicles, and two receivers synchronously tracking and observing the same GPS satellite. The pseudo-range observation value measured by the reference station is compared with the known distance, and the pseudo-range correction value is obtained, which is transmitted to the GPS receiver on the UAV through the data link. While observing the GPS satellites, the GPS receiver on the UAV receives the correction value from the reference station, corrects the observed pseudo-range, and then uses the corrected pseudo-range to perform positioning, and calculates the positioning parameters. The positioning result is WGS-84 Coordinates in a coordinate system between geocentric spaces.

A visual navigation system consists of an optical system (camera) and a video processor. The present invention adopts a monocular vision system, that is, the vision system has only one camera. The camera is fixed on the rotating platform, which allows the platform to rotate in two degrees of freedom, the azimuth angle and the pitch angle, so as to track the characteristic area, and mainly complete the characteristic feature collection of the landing area. Using an ordinary color camera, the pixels of the image taken are 1920×1080, and the signal-to-noise ratio is better than 48dB, which is used to convert the optical signal into a video signal and send it to the video processor.

The video processor includes a video capture card and an image processing circuit. The video capture card first amplifies the video signal output by the CCD, performs synchronous separation of field signals, and sends the detected video signal to the image processing circuit. The coordinates of the feature points in the image are obtained by processing the characteristic pattern of the landing area, and the coordinates are sent to the main control computer to solve the position parameters of the UAV relative to the landing area.

Digital image processing is the core of machine vision navigation system. The main process is divided into three steps: image preprocessing, feature point extraction and position calculation, as shown in Figure 2. The main control computer is the calculation and control center of the machine vision navigation system, responsible for data processing, after calculating the distance, height, and attitude angle of the UAV relative to the landing area, the data is transmitted to the UAV to control its positioning and landing. The UAV automatic driving system collects the parameters provided by the sensor according to the input control instructions, and generates control instructions according to the set control algorithm and logic, and realizes the control of the UAV by controlling the actuator.

Claims (5)

1. A method for positioning and landing of an unmanned aerial vehicle based on machine vision, comprising calculating positioning parameters, solving the position parameters of the unmanned aerial vehicle relative to the landing area and controlling the positioning and landing of the unmanned aerial vehicle; Calculate the positioning parameters: through two GPS receivers installed on the base station and the UAV, the same GPS satellites are tracked and observed synchronously; the pseudo-range observation value measured by the base station is compared with the known distance, and the pseudo-range correction value is calculated. It is transmitted to the GPS receiver on the drone through the data link; the GPS receiver on the drone observes the GPS satellite data, receives the correction value from the reference station, corrects the observed pseudo-range, and then uses the corrected pseudo-range for positioning. Calculate the positioning parameters; Solve the position parameters of the UAV relative to the landing area: complete the acquisition of the characteristic characteristics of the landing area through the camera fixed on the UAV; process the collected video signal, that is, the characteristic pattern of the landing area, and the video acquisition card will The video signal output by the CCD is amplified, and the synchronous separation of the field signal is carried out, and the detected video signal is sent to the image processing circuit; the image processing circuit processes the characteristic pattern of the shooting area to obtain the coordinates of the characteristic points in the image , send the coordinates to the main control computer to solve the position parameters of the UAV relative to the landing area; Control the positioning and landing of the UAV: After the main control computer calculates the distance, height and attitude angle of the UAV relative to the landing area, it transmits the data to the UAV; the UAV autopilot system collects The parameters provided by the sensor are used to generate control instructions according to the set control method and logic, and the drone is positioned and landed by controlling the actuator. 2. A kind of unmanned aerial vehicle positioning landing method based on machine vision as claimed in claim 1, is characterized in that, described camera is only provided with one, and its image capture pixel is 1920 * 1080. 3. a kind of unmanned aerial vehicle positioning and landing method based on machine vision as claimed in claim 1, is characterized in that, described positioning parameter adopts WGS-84 coordinate system. 4. A kind of unmanned aerial vehicle positioning and landing method based on machine vision as claimed in claim 1, it is characterized in that, described image processing circuit is to the characteristic pattern of photographing the landing zone and the specific method is to carry out grayscale successively Image correction, filtering, feature point extraction, non-feature point exclusion, rough registration, threshold selection, binarization, feature point extraction, feature point re-detection, linear pose calculation and result output. 5. a kind of unmanned aerial vehicle positioning and landing system based on machine vision, it is characterized in that, comprise following unit: the first GPS information receiving unit: the GPS receiver of described first GPS information receiving unit is installed on the base station, uses To compare the pseudo-range observation value measured by the reference station with the known distance, obtain the pseudo-range correction value, and transmit it to the GPS receiver installed on the drone through the data link; the second GPS information receiving unit: the second GPS The GPS receiver of the information receiving unit is installed on the UAV to observe the GPS satellite data, receive the correction value from the reference station, correct the observed pseudo-range, and then use the corrected pseudo-range to perform positioning and calculate the positioning parameters; Landing area acquisition unit: through the camera fixed on the UAV to complete the acquisition of the characteristics of the landing area; video signal processing unit: used to amplify the video signal output by the CCD, perform synchronous separation of field signals, and convert the detected video The signal is sent to the image processing unit; the image processing unit is used to process the characteristic pattern of the landing area of photography, obtain the coordinates of the feature points in the image, and send the coordinates to the main control computer to solve the UAV relative to the landing area. Position parameters; control command unit: used to calculate the distance, height and attitude angle of the UAV relative to the landing area, and then transmit the data to the UAV; landing execution unit: according to the input control command, the acquisition sensor provides parameters, and generate control commands according to the set control method and logic, and realize the positioning and landing of the UAV by controlling the actuator.